LoCo: Low-Bit Communication Adaptor for Large-scale Model Training

• URL: http://arxiv.org/abs/2407.04480v1
• Date: Fri, 5 Jul 2024 13:01:36 GMT
• Title: LoCo: Low-Bit Communication Adaptor for Large-scale Model Training
• Authors: Xingyu Xie, Zhijie Lin, Kim-Chuan Toh, Pan Zhou,
• Abstract summary: Low-bit communication often degrades training quality due to compression information loss. We propose Low-bit Communication Adaptor (LoCo), which compensates local local GPU nodes before, without compromising quality. Experimental results show that across moving large-scale training model frameworks like Megatron-LM and PyTorchs FSDP, LoCo significantly improves compression communication efficiency.
• Score: 63.040522637816906
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: To efficiently train large-scale models, low-bit gradient communication compresses full-precision gradients on local GPU nodes into low-precision ones for higher gradient synchronization efficiency among GPU nodes. However, it often degrades training quality due to compression information loss. To address this, we propose the Low-bit Communication Adaptor (LoCo), which compensates gradients on local GPU nodes before compression, ensuring efficient synchronization without compromising training quality. Specifically, LoCo designs a moving average of historical compensation errors to stably estimate concurrent compression error and then adopts it to compensate for the concurrent gradient compression, yielding a less lossless compression. This mechanism allows it to be compatible with general optimizers like Adam and sharding strategies like FSDP. Theoretical analysis shows that integrating LoCo into full-precision optimizers like Adam and SGD does not impair their convergence speed on nonconvex problems. Experimental results show that across large-scale model training frameworks like Megatron-LM and PyTorch's FSDP, LoCo significantly improves communication efficiency, e.g., improving Adam's training speed by 14% to 40% without performance degradation on large language models like LLAMAs and MoE.

Related papers

• Accelerating Large Language Model Training with Hybrid GPU-based Compression [3.204387803072905]
  MPI libraries have been proven to reduce message size significantly and leverage interconnect bandwidth. We investigate the efficacy of compression-assisted MPI collectives under the context of distributed Large Language Model (LLM) training.
  arXiv  Detail & Related papers  (2024-09-04T04:05:30Z)
• Accelerating Communication in Deep Learning Recommendation Model Training with Dual-Level Adaptive Lossy Compression [10.233937665979694]
  DLRM is a state-of-the-art recommendation system model that has gained widespread adoption across various industry applications. A significant bottleneck in this process is the time-consuming all-to-all communication required to collect embedding data from all devices. We introduce a method that employs error-bounded lossy compression to reduce the communication data size and accelerate DLRM training.
  arXiv  Detail & Related papers  (2024-07-05T05:55:18Z)
• Activations and Gradients Compression for Model-Parallel Training [85.99744701008802]
  We study how simultaneous compression of activations and gradients in model-parallel distributed training setup affects convergence. We find that gradients require milder compression rates than activations. Experiments also show that models trained with TopK perform well only when compression is also applied during inference.
  arXiv  Detail & Related papers  (2024-01-15T15:54:54Z)
• GraVAC: Adaptive Compression for Communication-Efficient Distributed DL Training [0.0]
  Distributed data-parallel (DDP) training improves overall application throughput as multiple devices train on a subset of data and aggregate updates to produce a globally shared model. GraVAC is a framework to dynamically adjust compression factor throughout training by evaluating model progress and assessing information loss associated with compression. As opposed to using a static compression factor, GraVAC reduces end-to-end training time for ResNet101, VGG16 and LSTM by 4.32x, 1.95x and 6.67x respectively.
  arXiv  Detail & Related papers  (2023-05-20T14:25:17Z)
• Communication-Compressed Adaptive Gradient Method for Distributed Nonconvex Optimization [21.81192774458227]
  One of the major bottlenecks is the large communication cost between the central server and the local workers. Our proposed distributed learning framework features an effective gradient gradient compression strategy.
  arXiv  Detail & Related papers  (2021-11-01T04:54:55Z)
• Quantization for Distributed Optimization [0.0]
  We present a set of all-reduce gradient compatible compression schemes which significantly reduce the communication overhead while maintaining the performance of vanilla SGD. Our compression methods perform better than the in-built methods currently offered by the deep learning frameworks.
  arXiv  Detail & Related papers  (2021-09-26T05:16:12Z)
• Towards Compact CNNs via Collaborative Compression [166.86915086497433]
  We propose a Collaborative Compression scheme, which joints channel pruning and tensor decomposition to compress CNN models. We achieve 52.9% FLOPs reduction by removing 48.4% parameters on ResNet-50 with only a Top-1 accuracy drop of 0.56% on ImageNet 2012.
  arXiv  Detail & Related papers  (2021-05-24T12:07:38Z)
• ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training [74.43625662170284]
  Large-scale distributed training of Deep Neural Networks (DNNs) on state-of-the-art platforms is expected to be severely communication constrained. We propose a new compression technique that leverages similarity in the gradient distribution amongst learners to provide significantly improved scalability. We experimentally demonstrate that ScaleCom has small overheads, directly reduces gradient traffic and provides high compression rates (65-400X) and excellent scalability (up to 64 learners and 8-12X larger batch sizes over standard training) without significant accuracy loss.
  arXiv  Detail & Related papers  (2021-04-21T02:22:10Z)
• An Efficient Statistical-based Gradient Compression Technique for Distributed Training Systems [77.88178159830905]
  Sparsity-Inducing Distribution-based Compression (SIDCo) is a threshold-based sparsification scheme that enjoys similar threshold estimation quality to deep gradient compression (DGC) Our evaluation shows SIDCo speeds up training by up to 41:7%, 7:6%, and 1:9% compared to the no-compression baseline, Topk, and DGC compressors, respectively.
  arXiv  Detail & Related papers  (2021-01-26T13:06:00Z)
• Sparse Communication for Training Deep Networks [56.441077560085475]
  Synchronous gradient descent (SGD) is the most common method used for distributed training of deep learning models. In this algorithm, each worker shares its local gradients with others and updates the parameters using the average gradients of all workers. We study several compression schemes and identify how three key parameters affect the performance.
  arXiv  Detail & Related papers  (2020-09-19T17:28:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.